New technologies can help businesses and governments make better decisions. A panel of experts looks at how to break down barriers to adoption.
|Make international payments faster and cheaper with World Account from WorldFirst.|
|Sign up today – and start saving!|
Seated at this virtual roundtable are James Hodge, global chief technical adviser, Splunk; Jessica Lennard, senior director, global data and artificial intelligence initiatives, Visa; Tom Symons, head of government innovation research, Nesta; Roger Taylor, chair of the Centre for Data Ethics and Innovation; and Siân Thomas, chief data officer, Department for International Trade.
Q Is there proof that better use of data really feeds through to the bottom line for the private sector?
JL If you use data in an unethical or irresponsible way, you’re going to destroy trust. But if you do it right, I think the proof is there. At Visa, we have a neural network-based product that monitors for fraud in real time and does a risk-based analysis. That AI model alone prevents $25 billion of fraud a year for consumers and businesses. That’s pretty good proof.
JH For all businesses, it’s imperative to transform, and they need to give the people on the frontline who make decisions the right data and context. We released some research earlier in the year showing data innovators, organisations that put a data strategy at the front of everything they do, generate a higher percentage of revenue from new product lines. This shows innovation driven by data impacts the bottom line. We’re at what I would call the dawn of the data age. If we can get it right, then the way we can accelerate companies and economies over the next five to ten years is going to be breathtaking.
Q So what is holding back the use of these transformative data technologies?
RT At the Centre for Data Ethics and Innovation, we published an AI barometer that identifies barriers to data-driven innovation across various sectors. These include low data quality and availability; lack of co-ordinated policy and practice; lack of transparency around AI and data use; and lack of knowledge about the real impacts of new technologies.
JL I think it is incredibly important to build awareness. Good use of data brings value to the public, value to businesses, value to society more broadly. But we need to make that case, we need to demonstrate those good outcomes.
Q Is there a tremendous amount of valuable data that is unused?
JH We often refer to that data as “dark data”, nothing nefarious, but just data that is not being used for competitive advantage. One reason is that people think it’s hard to embark on a data project. People think there has to be a technological answer to everything, but actually it’s much more about collaboration.
JL The aim is not just to collect and generate as much raw data as you possibly can; that’s not going to build trust. You need to be innovating with purpose, especially if you’re talking about personal data.
Q There’s a vision of data priests handing down tablets of statistical-driven wisdom to the rest of the organisation. Is that what happens?
JH Actually, that’s completely the wrong way to do it.
ST If you put a team of data scientists in a room and leave them there for six months, they will come up with some fantastic things, but then that needs to be translated into the real world. You need to work with the frontline staff, the people who can use those insights, as well as more senior people to make sure the strategic direction of what you’re doing is appropriate. Cultural sensitivity and working collaboratively across a whole range of different professions is very beneficial.
TS To have a data-informed organisation is all about a huge cultural shift in terms of how the organisation sees data and how it uses it. You can spend a lot of time doing the technical work to build a great data product, but if you’re not investing enough of that time understanding how it’s going to fit into someone’s day-to-day job, educating about it, informing them about it, then you might miss a trick.
Q So do ideas for advanced data use come bottom up as well as top down?
ST In our world [government], they certainly do come from the bottom up as well as the top down. Quite often, the top-down approach has a lot of strategic oversight, but they don’t have the detail about how things work. Recently we created a data ethics board to invite different people into the conversation about data. It’s only by working from the bottom up that you manage to implement those technologies in a way which supports the business and delivers on its strategy.
Q Is there a problem understanding data at the highest decision-making levels?
RT Organisations are already data driven and the people at the top of organisations know how to understand the data presented to them, such as a financial statement. What they don’t yet understand is how to read information about complex, data-driven systems. That information, just like a balance sheet or set of accounts, is never going to tell you directly what your right strategy is as a company. But if you can’t interpret it, then you will not be able to manage the AI-driven organisation of the future effectively. So this is a skill that will need to be developed.
JL It’s not necessarily the data itself the board needs to understand; it’s the insights. And the board should be using human-centric thinking. They should be saying, “If I can’t understand the insights here, I shouldn’t be relying on them too much.”
Q The leadership of a financial services group would be expected to be data literate, but how does it work in government?
TS We’re not asking senior leaders in the public sector to fire up their laptop and do some coding, but they need to understand the basics and know what questions to ask. That may be a level of data literacy that has to be grown.
ST I think, for the most part, there is data literacy within government departments. And where the individuals making those decisions don’t have the skills personally, then there are processes in place that make sure, for example in my department, the chief statistician or me as chief data officer is invited to meetings to contribute.
Q Are there different types of data-driven decision-making?
TS At Nesta we refer to the goal as data-informed decision-making, rather than data-driven. Governance and ethics are really important and to ensure those principles are adhered to, it’s important to have a human at the end of the decision chain.
RT We sometimes pretend you can always put a human in the loop, but when it comes to what Spotify recommends, or a self-driving car slamming on the brakes, there clearly won’t be. This is a really significant area that requires clarification; we have to understand exactly when decisions can be automated, when is it OK or even necessary?
For more information please visit splunk.com
Excerpt from Raconteur Business Growth & Recovery special report published in The TIMES.
|Want more great content and market-leading insight and analysis sent to your inbox?|
|Sign up for our newsletter|